Outcome-Based Context Engineering

Context is architecture. Outcomes are verification.

I don't engineer prompts. I engineer the environment in which optimal outcomes become inevitable — constraint systems, decision traces, and verification mechanisms for AI-native product organisations.

Open to Governor-level engagements
The Four Laws
I

Outcome Primacy

The Outcome must be defined before the Context. Context is fuel — without a destination, you cannot choose the right fuel.

II

Context Sanctity

Signal-to-noise ratio determines intelligence. The context window is finite. Irrelevant context degrades reasoning.

III

Constraint

An unconstrained model is a hallucination machine. Performance is maximized by aggressively reducing the search space.

IV

Verification

Trust is not a metric. If the outcome cannot be verified, the context is invalid.

Decision Trace Lifecycle
Intent Manifest
→
Context Prompt
→
Execution Model
→
Verification Outcome
What I deliver
Context Architecture

Design the knowledge environment

Knowledge graphs, constraint systems, decision traces. Making AI systems predictable and auditable.

Outcome Engineering

Define before you build

Verifiable outcomes before system design. Validation mechanisms that prove success deterministically.

Model Orchestration

Route decisions intelligently

Right model for the right task. Complexity-aware routing based on cost, traces, and historical data.

Verified outcomes
Context Graph

JobEasy professional knowledge graph

Graph-first domain model with decision traces and provenance tracking. Explainable matching with temporal context awareness.

✓
Constraint

Robotics abstraction framework

Hardware-agnostic integration with aggressive negative constraints across ROS2, MuJoCo, IsaacSim. Deterministic behaviour.

✓
Trace Capture

Prompt-engineered UX generation system

Reusable context templates with embedded verification hooks. Consistent quality with full decision lineage.

✓
LLM Product

AI-generated commerce pipeline

Automated creative generation with commercial validation constraints. Design-to-market with verifiable quality gates.

✓
Manifest

Camping platform disruption framework

South African market-entry with outcome-first constraints. Defensible wedges with verifiable go/no-go criteria.

✓
Analysis

AI coding platform benchmarking

Platform-thinking analysis with traceable evaluation criteria. Auditable scoring methodology for positioning.

✓
Governance

PAIA compliance architecture

South African regulatory framework with risk mitigation constraints. Executive-level governance with verifiable compliance trails.

✓
Cost Route

AI model pricing analysis

Unit economics verification with founder-level resource constraints. Cost-optimized routing with measurable efficiency.

✓
Thinking
Prompt engineering is art. Context engineering is science. 12 min
The two clocks problem: why event time matters more than state time 9 min
From manager of people to manager of agents: the context curator role 14 min
About

I architect context environments, not just products.

OBCE is the methodology I developed and operate by. It treats AI not as a tool to be prompted, but as a system to be constrained. Every project I take on begins with a Manifest — the defined outcome — and ends with deterministic verification.

Based in South Africa with global reach. I operate as a Level 3 Governor: managing context graphs, analysing decision traces, and building the organisational world models that make complex AI-native products succeed.

  • L1 Operator — Implement OBCE tools, pass audit scoring
  • L2 Architect — Define manifests, design constraint systems
  • L3 Governor — Manage context graphs, organisational world models

I take on engagements that require Governor-level thinking: context architecture, decision trace systems, and organisational AI strategy.

Start with your Manifest →